Deciphering the Complex Communication Networks That Orchestrate Pancreatic Islet Function

Cells have an innate ability to form tissues. Indeed, within the pancreas, the endocrine cells responsible for maintaining glucose homeostasis coalesce during development into small organs called islets. In the islet, the insulin-secreting β-cells make direct cellular contacts with other β-cells as well as with other endocrine cell types, including the glucagon secreting α-cells and the somatostatin-secreting δ-cells. The importance of these cellular contacts for islet cell physiology was first reported nearly 30 years ago when Halban, Meda, Wollheim, Weir, Pipeleers, and colleagues showed that hormone secretion is diminished in dispersed cells compared with intact islet clusters (1–3). Results from these creative experiments showed that β-cells in islets roughly secrete 18 times more insulin than single β-cells and that β-cell pairs secrete four to five times more insulin than expected from their sum (1,3,4). Thus, cooperation between β-cells in the islet is not additive but highly synergistic.

This arithmetical anomaly poses a significant problem to the investigator: the complexity of cellular interactions in the tissue makes it difficult to comprehend organ function. To properly perform their homeostatic functions, these specialized endocrine cells have developed intricate cell-cell communication mechanisms that rely on the recognition of the state of activity of their neighboring cells. These communication mechanisms include direct contact signaling with neighboring cells and with the extracellular matrix (gap junctions, ephrins, cadherins, and integrins) as well as noncontact signaling through secretion of paracrine factors (5). Loss of these communication mechanisms has a profound effect on the secretory function of endocrine cells in the pancreatic islet. This article discusses current developments in the field of islet endocrine cell cross talk, emphasizing novel analytical and synthetic approaches that are overcoming major roadblocks in studying communication networks in the islet.

When dispersed, islet cells have the urge to reaggregate (6), and once they have formed clusters they secrete more insulin in response to glucose (3). To understand the sociobiology that makes endocrine cells work more efficiently in clusters, islets can be examined using analytical tools, that is, by manipulating and imaging individual components and measuring the impact on the whole islet. In this section, we discuss recent developments that are helping define intraislet signaling by using analytical approaches. The research work we describe employs two different strategies to decipher the complexity of islet communication, namely, dissecting out individual components of cell-cell communication and multicellular imaging of islet networks.

There are three basic types of communication between islet cells: electrical coupling through gap junctions, direct cell-to-cell contacts, and paracrine interactions. Paracrine signaling involves release of a chemical signal, which travels through the local microcirculation or interstitial fluid to reach its target cell. The list of putative signaling molecules in the islet keeps increasing (e.g., urocortin 3 [7]). The emerging importance of paracrine signaling from α- and δ-cells in orchestrating insulin secretion was reviewed recently (7,8). Here, we focus on new findings related to the paracrine roles of purinergic signals (ATP and adenosine) and GABA because they illustrate the versatility and intricacy of cell-cell communication in the islet.

ATP is known to be stored in insulin granules and is thus cosecreted with insulin. Interestingly, ATP may leave the granule before or even without insulin in what is known as kiss-and-run exocytosis. This ATP first encounters and binds to receptors on β-cells, thus triggering signaling cascades that amplify insulin secretion from human islets (Fig. 1). Initially, these effects were reported to be mediated by ionotropic P2X₃ purinergic receptors (9), but newer studies revised this picture to include metabotropic P2Y1 receptors (10,11). ATP affects other endocrine cells as well (12). In its journey through the interstitium, ATP also faces the resident macrophage. Macrophages in general are endowed with purinergic receptors, and indeed two recent studies from our group show that macrophages in mouse and human islets are exquisite sensors of extracellular ATP (13,14). Of note, the results indicate that islet resident macrophages partially depend on endogenous ATP input from β-cells to produce and secrete cytokines and metalloproteinases. Furthermore, purinergic receptor expression in islet macrophages was found to be downregulated in obese and diabetic states. In these states, the loss of ATP sensing in macrophages may reduce their secretory capacity and impair their interactions with the microenvironment.

As soon as ATP starts percolating through the interstitial space it is cleaved by potent ATPases that generate other purinergic signals such as ADP (15). ADP may be further cleaved to adenosine once it approaches ectonucleotidase-coated capillaries. This adenosine likely inhibits the activity of vascular pericytes, thus dilating capillaries and increasing local blood flow, as shown recently (16). These effects are mimicked by stimulation of β-cells with high glucose concentrations and can be blocked with an A1 receptor antagonist (16). These novel findings demonstrate how purinergic signals released from β-cells accomplish multiple tasks in the islets: 1) by potentiating insulin secretion, ATP increases the speed and robustness of the β-cell’s early response to a rise in glycemia; 2) by signaling to macrophages, ATP conveys information about the β-cell’s secretory status, therefore allowing macrophages to properly adjust tissue homeostasis; and 3) by inhibiting pericytes, adenosine regulates islet blood flow to deliver insulin more efficiently. Thanks to their versatility, purinergic signals are thus able to recruit multiple cell types that help orchestrate hormone secretion and maintain islet integrity (Fig. 1).

GABA has been postulated as a paracrine signal since it was reported to inhibit α-cells (17). Recent studies characterized in detail native, high-affinity GABA_A receptors in human β-cells (18). Because GABA also inhibits immune cells, it was further suggested that it could protect β-cells under immune attack (19). In 2017, a study appeared describing GABA as an inducer of α-to-β-cell conversion in vivo, which was presented as an unprecedented hope toward improved therapies for diabetes (20). The results were sensational and triggered a research mania. However, efforts at reproducing these results failed, as long-term treatment of mice and rhesus monkeys with GABA (or artemisinins) did not cause α–to–β-cell transdifferentiation (21,22). In brief, GABA may be famous as a paracrine signal but infamous as a transdifferentiating factor.

It was interesting to follow this debate, in particular because we were aware that it had not been established unequivocally how GABA is secreted in the islet. Like other paracrine signals, islet GABA release was described to depend on exocytosis, although researchers suspected a large tonic glucose-independent background release of GABA (23). A recent study now shows that the β-cell effluxes GABA from a cytosolic pool in a pulsatile manner (24) (Fig. 2). Moreover, knocking down expression of subunits for volume-regulated anion channels (VRACs) abolished GABA release, indicating that VRACs are critical for GABA secretion in the islet. GABA content in β-cells is depleted and secretion is disrupted in islets from patients with type 1 and patients with type 2 diabetes, suggesting that loss of GABA correlates with diabetes pathogenesis (24). It still has to be established whether this constitutive GABA secretion promotes the oscillatory activity pattern of the islet and whether its loss contributes to the erratic oscillations in insulin secretion seen in type 2 diabetes. In several ways, the notion that GABA in the islet behaves as an organic osmolyte whose extracellular levels oscillate periodically is highly unconventional. More than 30 years after the seminal work of Rorsman et al. (17), many aspects of GABA signaling in the islet remain mysterious and deserve further investigation.

A different analytical approach to reveal communication networks is to study multicellular activity in intact islets. In the last decade, different groups have examined coordinated responses in islets using complex microfluidic devices (25,26), pancreas slices (27), in vivo imaging of intraocular islet grafts (28), or intravital imaging in Zebrafish (29). Mathematical modeling is then applied to the data sets for understanding of how individual events produce episodic hormone responses (e.g., multicellular islet models [30]). The new analytical approaches are showing that β-cell collectives work as broadscale complex networks and share similarities in global statistical features and structural design principles with the internet and social networks (29,30).

These studies are revealing properties of the islet network that are only apparent if the population is examined as an ensemble. A common thread appears to be β-cell heterogeneity. Physiologists have been aware for decades that the β-cell and other endocrine cell populations are heterogenous (31), a feature that was only recently confirmed at the molecular level (reviewed in 32). Heterogeneity, a natural trait of most cell collectives, introduces robustness and plasticity—for instance, different β-cell response patterns to glucose allow insulin secretion to be fine-tuned. That some β-cells recover from stress while others continue secreting reduces the islet’s vulnerability to insults. Using high-speed multicellular Ca²⁺ imaging combined with correlation analyses, Hodson and colleagues (33) found rare superconnected hub cells whose activity tends to precede that of the remainder of the population. These studies provide evidence that pacemaker-like β-cells exist in the islet. Thus, some β-cells may be more equal than others.

A shortcoming is that each analytical modality seems to identify its own type of heterogeneity, and it is difficult to find biological correlates. Another caveat of these multicellular recordings is that they rely heavily on data obtained with indicators for bulk cytoplasmic Ca²⁺ concentrations. This Ca²⁺ signal represents the sum of multiple inputs, has temporal features that reflect the indicator’s Ca²⁺ binding kinetics, and, unlike local Ca²⁺ influx through particular Ca²⁺ channels, is not necessarily coupled to secretory output. The imaging techniques to assess islet signaling are not fast enough to measure the primary electrical signal of the islet, the action potential. Fluorescent imaging can only predict electrical connectivity of an islet. Microelectrode array–based technologies or the use of modern optical probes for membrane voltage may provide a clearer assessment of connectivity. This is important because examining discrete pulsatile secretory events in intact human islets did not reveal leaders and followers (34) (discussion of the hub cell model in 35,36). Molecular biology data suggest that β-cell subpopulations transition through different states (stressed vs. active) multiple times over the course of their life span, suggesting that there are no fixed populations (32). Also, an exposed dominant cell would render fragile a system whose robustness is based on ensemble activity. But maybe that is how diabetes can ensue. These comments notwithstanding, it is clear that multicellular imaging studies coupled to mathematical modeling are revealing emergent properties of islet communication networks that will complete our model of islet biology and explain its fragility.

Mathematical modeling using data from multicellular imaging may be a first step in providing a synthesis, but in general the reductionist (analytical) approach identifies individual components in a biological system without characterizing the system’s entirety. Synthetic approaches can be used to overcome this limitation by combining a known set of isolated components to reconstruct the system and understand its complexity. Researchers can thus recreate miniaturized or simplified versions of an organ in vitro. These so-called organoids self-organize and reassemble in ways that mimic the original tissue composition. Importantly, they recapitulate specific functions of the organ (Fig. 3).

Reverse engineering is the process by which a man-made object (in engineering) or an organ (biomedical research) is deconstructed to reveal its design or architecture or to extract knowledge from the object or organ. By deconstructing and reassembling components, an engineer can determine the contribution of each of the individual components. Engineers have used this approach for centuries to build bridges, cars, and computers. Islet biologists have applied the same approach by reconstructing the individual cellular components of islets to create islet-like organoids, also known as pseudoislets. These pseudoislets have been created from cell lines, primary cells, or stem cells. Recent advances in islet cell-cell communication based on the use of pseudoislets are discussed here. Before we delve more into the subject, it is important to note that pseudoislet formation does not always produce the architecture of the islet cells from which the pseudoislets were made. This could impact paracrine signaling, hub cell or leader cell contacts, and many other signaling events within the islet.

Using islet hormone-secreting cell lines for islet research is an attractive approach that has numerous advantages: they can be propagated in culture and are easily transformable and cost-effective. Cell lines such as MIN6 are generally maintained as adherent monolayers in tissue culture but, when subcultured in tissue culture flasks precoated with gelatin (1%), develop into three-dimensional pseudoislets (4). MIN6 cells also form pseudoislets when cultured in dishes without tissue culture treatment and can be formed with suspension culture (for example, see 37). For human pseudoislet formation, innovative plates with microwells are required. Like primary islets, pseudoislets depend on cell adhesion molecules such as E-cadherin and N-cadherin to maintain their structure (4,38). Pseudoislets have been made from numerous islet hormone-secreting cell lines including the insulin-secreting MIN6 cells (4), glucagon-secreting αTC, and somatostatin-secreting TGP-52 cells (39,40).

Interestingly, reaggregation of αTC and MIN6 cells induced spontaneous organization of the rodent islet core-mantle structure (6,41,42). These findings suggest an intrinsic hardwiring in the islet endocrine cytoarchitecture. Studies suggest cell-cell contact mechanisms such as those mediated by Robo receptor signaling may control the cytoarchitectural arrangement (43). There is no doubt that studies using cell lines have revealed fundamental mechanisms of insulin secretion (44). However, immortalized cells are highly proliferative. Proliferating β-cells lose the phenotype typical of β-cells and secrete less insulin than their differentiated counterparts. Most primary β-cells, by contrast, are fully differentiated and can be as old as the animal in which they reside (45). Nonimmortalized cells such as primary islets or terminally differentiated stem cells thus seem more suitable for studying islet physiology.

Primary islets have been the workhorse for diabetes research since 1967, when the islet isolation protocol was established (46). Since then, numerous fundamental communication mechanisms have been discovered with use of intact islets. Although these findings contribute to our understanding of islet biology, using a reverse engineering approach by dismantling primary islets and reconstructing the cellular pieces back together offers researchers a unique discovery tool. One advantage of using pseudoislets derived from primary cells compared with native intact islets is that they can be readily manipulated genetically (47,48). These studies have opened the door for new functional studies in human islets using delivery of viral constructs including designer receptors exclusively activated by designer drugs (DREADDS [49]), overexpression (47), and knockdown studies using shRNA (50). A further advantage is that insulin and glucagon secretion are maintained in human pseudoislets over a longer culture period (51). In addition to traditional methods of obtaining islets, pseudoislets derived from primary cells can now be commercially obtained as well. Companies such as InSphero AG (Schlieren, Switzerland), and likely others in the future, are helping to facilitate research. The techniques described above may prove useful for engineering islets that result in a better clinical success in the context of therapeutic transplantation of islets into patients with diabetes.

Human embryonic stem cells and induced pluripotent stem cells (iPSCs) can now be differentiated into β-like cells and matured into pseudoislet organoids (52,53). Because of the proliferative capacity of organoid clusters, their production from stem cells offers a unique solution to a major challenge in the field of human islet transplantation, namely the paucity of islet donors. Although these organoid clusters often contain polyhormonal cells that have defects in glucose-stimulated insulin secretion and lack expression of key β-cell–specific transcription factors (54,55), recent improved protocols of stem cell differentiation into β-like cells have shown that these cells secrete insulin in response to glucose and improve glucose tolerance after transplantation into diabetic mice (56). Of note, an important step of this protocol is to cluster and resize iPSC-derived β-cell pseudoislets (∼360 μm) into pseudoislets that span 170 μm after resizing. This resizing step improved insulin secretion by nearly 25%. Interestingly, this size is close to the typical islet size in nearly all vertebrate species (100–200 μm diameter) despite the wide range of total pancreatic volume. Thus, there seems to be a limit to how beneficial cell crowding is and there may be an ideal cell number for concerted cell-cell communication. While hypoxia contributes largely to islet size limitation in vitro (57), the mechanisms that determine islet organ size in vivo still require investigation.

Perhaps the most intriguing applications for stem cell–derived pseudoislets have yet to reach their full potential. Pseudoislets derived from donor stem cells of donors with diabetes may provide a feasible means to deliver personalized medicine to patients, allowing exploration of the mechanisms causing β-cell failure in the particular donor. A crucial step toward reaching this goal will be to develop reliable systems for modeling diabetes. Recently, several studies have reported generating β-like cells from human pluripotent stem cells from donors with type 1 diabetes (T1D) (58). In addition to endocrine cells, iPSCs from donors have been used to generate immune cells from T1D patients (59). In the future, a complete model for diabetes pathogenesis would not only include T1D donor endocrine cells derived from iPSCs but would additionally include other cell types that may play a role in diabetes pathogenesis. These studies may help overcome another major roadblock for therapeutic islet transplantation: the current need for immunosuppression. Whereas immunosuppression concerns may be addressed by encapsulating of organoids (60), it should be feasible to manipulate cell-surface signaling and the anti-inflammatory transcriptional machinery to make islet organoids less immunogenic.

It should be noted that pseudoislets may not necessarily be less physiological than cultured primary islets. Isolation of pancreatic islets from their natural environment is a traumatic event that requires exocrine tissues to be digested and islets to be rescued from this debris. Not surprisingly, islet isolation changes gene expression patterns in the islet (61). The cytoarchitecture may be maintained in isolated islets (i.e., core/mantle structure in mouse islets), but many cellular components that are important for islet function are lost during extended culture. These cells include immune cells (62), endothelial cells (63), and nerves (64). Future protocols to produce hybrid pseudoislets consisting of several cell types may more faithfully reproduce the complex natural anatomy of the islet. This section will discuss how incorporating these critical cell types in pseudoislet organoids may improve our understanding of islet physiology and diabetes pathophysiology.

Islet resident macrophages are emerging as important players in β-cell proliferation (65,66), sensing of β-cell activity (13,14), islet inflammation (67), and antigen presentation (68). Therefore, macrophages should be considered in future pseudoislet studies. These studies may reveal basic physiological questions that remain unanswered, such as, what are the local signals that recruit resident macrophages to the islet or how do islet immune cells respond to therapeutic intervention? Addressing these questions using pseudoislets in vitro or after transplantation may lead to therapeutic advances in diabetes through immunomodulation.

Although tissue-resident immune cells may proliferate locally (69), recruitment of immune cells also plays an important role in type 1 and type 2 diabetes (70,71). The diverse repertoire of immune cells found in the islet during diabetes includes natural killer cells, cytotoxic T cells, B cells, and macrophages (72). A large proportion of these immune cells are recruited to the islet through the islet microvasculature. Mixed lymphocyte islet culture allows studying activation of responding lymphocytes, cytokine secretion, and changes in islet cell antigen expression (73). Coculture of diabetogenic CD4 and CD8 T cells with islet showed that these immune cells secrete soluble factors promoting β-cell proliferation, suggesting that infiltrating may have surprising therapeutic effects (74). We are not aware of similar studies that evaluate the impact of immune cells using pseudoislets. While these studies highlight how soluble factors from immune cells affect islet function, a properly functioning vasculature network is needed to model the recruitment, infiltration, and extravasation of immune cells into the pseudoislet as well as the direct interactions between immune cells and β-cells.

The specialized network of capillaries in the islet allows for efficient transport of micronutrients to pancreatic endocrine cells and of islet hormones out of the islet. Building and maintaining the islet vasculature require proper coordination between vascular cells, stromal cells, and endocrine cells of the islet parenchyma. In addition to the factors that maintain the vasculature in the long-term (e.g., VEGF), islet blood flow can be acutely modulated by local signals derived from β-cells (adenosine) and by neural input from sympathetic nerves (16). There is evidence that the islet microcirculation is dysfunctional during diabetes (75,76), as well as after islet transplantation (77). By incorporating the islet vasculature into pseudoislet organoids, studies could help improve graft revascularization after islet transplantation. Moreover, as secretion is polarized toward the vasculature (78), a “vascularized” pseudoislet would provide a more physiological model for study of hormone release. Moreover, reincorporation of vascular cells will likely contribute to build the extracellular matrix of the islet organoid.

Vascularized organoids have helped investigators gain insight into physiology and pathophysiology in other research areas including liver (79), lung (80), and brain (81). Therefore, methods to deliver endothelial cells to pseudoislets would be a powerful tool for future islet research. Indeed, mosaic pseudoislets can be generated in vitro by mixing of dispersed islets with endothelial progenitor cells (82). This approach can be extended by transplantation of mosaic pseudoislets and determination of how the graft’s structure and function recapitulate the features of islets in the pancreas (83).

The pancreatic islet is innervated by parasympathetic, sympathetic, and sensory nerves (64). While the innervation density and target cells may vary between species (64), the signaling molecules they release (e.g., acetylcholine and norepinephrine) have been shown to regulate islet blood flow (16) and promote insulin secretion in both mouse and human islets (84,85). The loss of sympathetic innervation seen in T1D may result in a substantial impairment of glucagon responses (86). Inclusion of neurons in organoid culture may enable investigating the neural defects observed in the islet in autoimmune diabetes.

How can islet innervation be engineered in vitro? Given that vascular endothelial growth factor (VEGF) coordinates islet innervation via vasculature scaffolding (87), it is likely that a pseudoislet organoid will need a functional vasculature before inclusion of neuronal cells in a pseudoislet organoid. Neurons could be added to a pseudoislet mix of endocrine and vascular cells. This would mimic the close association that neuro-insular complexes have with the peripheral vasculature of the islet. An alternative is to coculture neuronal organoids with pseudoislets. This would allow exploring the factors that attract neuronal projections to the islet and studying how innervation affects islet function. Here, the classical assay of Nobel Prize winner Rita Levi-Montalcini comes to our minds: a sensory ganglion is dissected out, cultured, and examined for nerve fiber outgrowth after exposure to different factors. This approach has been used recently to show that β-cells migrate toward explants of sympathetic ganglia, thus demonstrating that nervous input provides inductive cues for islet architecture (88). One can only imagine the myriad of possible combinations of different types of ganglia with pseudoislets of varying compositions. Of note, this approach can be extended to the in vivo situation by cotransplantation of two types of organoids into the anterior chamber of the eye, as has been done for tissues from different brain regions (89).

The decoding of paracrine signaling in the islet has significantly increased our understanding of the biology of this important micro-organ. The rate of new discoveries in islet research is at an all-time high. As we were writing this article, two new interesting articles on islet organoids were published (90,91). Hence, it is hard to reconcile all these findings and fit them into a cohesive physiological model. We have discussed studies using tried-and-tested methodologies as well as new analytical and synthetic approaches and mentioned the discoveries that they generated. We propose that more research with pseudoislets is needed to decipher complex interactions between endocrine, vascular, nerve, and immune cells. Generating mosaic islet organoids will be crucial for creating disease-modeling platforms aimed at revealing mechanisms of diabetes pathogenesis. While the source of tissue material in pilot studies will likely be renewable cell sources, such as cell lines, we advocate the use of human islets when possible, given their increased availability and relevance for the human disease. The studies discussed in this review have provided insight into the mechanisms of intraislet communication and how they impact insulin secretion. Knowing how interactions between different cell types change under stress conditions could inform approaches to increase or restore β-cell function in diabetes.

Funding. The authors’ work was supported by the Diabetes Research Institute Foundation and National Institutes of Health grants R56DK084321 (A.C.), R01DK084321 (A.C.), R01DK111538 (A.C.), R01DK113093 (A.C.), U01DK120456 (A.C.), R33ES025673 (A.C.), and R21ES025673 (A.C.); Leona M. and Harry B. Helmsley Charitable Trust grants G-2018PG-T1D034 and G-1912-03552; and the American Heart Association 19POST34450054 (J.W.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.